Pseudopotential-based Density-Functional Theory (DFT) permits the calculation of material properties with a modest computational effort, besides an acknowledged tradeoff of generating and testing pseudopotentials that reproduce established benchmark structural and electronic properties. To facilitate the needed benchmarking process, here we present a pragmatic method to optimize pseudopotentials for arbitrary materials directly from eigenvalue sets consistent with all-electron results. This method thus represents a much needed pragmatic route for the creation and assessment of sensitive pseudopotentials for DFT calculations that has been exemplified within the context of the SIESTA code. Comprehensive optimized pseudopotentials, basis sets, and lattice parameters are provided for twenty chemical elements in the bulk, and for both LDA and GGA exchange–correlation potentials. This method helps addressing the following issues: (i) the electronic dispersion and structural properties for Ge, Pd, Pt, Au, Ag, and Ta better agree with respect to all-electron results now, (ii) we provide the expected metallic behavior of Sn in the bulk – which comes out semiconducting when using available pseudopotentials, (iii) we create a validated pseudopotential for LDA-tungsten, and (iv) we create the first Bi pseudopotential for SIESTA that reproduces well-known electron and hole pockets at the L and T points. We investigated the transferability of these pseudopotentials and basis sets, and predict a new phase for two-dimensional tin as well.

This paper was originally published in Computational Materials Science, 98 (2015), Pages 372–389.

Download full text pdf of article